Use of the ZELIG Forest Simulation Model in the Uses of ...lands, forest structure may be more...
1
Use of the ZELIG Forest Simulation Model in the Coastal Landscape Analysis and Modeling Study (CLAMS) Robert J. Pabst 1 , Matthew N. Goslin 2 , Thomas A. Spies 3 , and Steven L. Garman 1 1 Department of Forest Science, OSU 2 NOAA, Santa Cruz, CA 3 USDA Forest Service PNW Research Station, Corvallis Introduction Background Simulating the development and distribution of forest types across the landscape is a key component of CLAMS, requiring robust models of stand-level dynamics. Two forest growth models are being used in CLAMS. ORGANON v.6.0 (Hann et al. 1997), an empirical growth and yield model, is used to simulate stands under private and tribal land ownerships. Stands on federal and state lands are modeled with ZELIG.PNW v.2.0 (Urban et al. 1993, Garman et al., in press). ZELIG is a gap model that can simulate a wide range of stand conditions and processes, thus it should be better suited than ORGANON to modeling forests on public lands where, in comparison to private lands, forest structure may be more complex, forest reserves exist, and alternative silviculture systems are often used. In this poster we present how we modified ZELIG for the Coast Range, how it performed with independent data sets, and its uses in CLAMS. Model structure of ZELIG ZELIG models the forest as a collection of gaps, with each gap representing the area potentially occupied by a canopy dominant tree (Urban and Shugart 1992). The model simulates each gap as it progresses through different phases of development, tracking all of the individual trees within it. ZELIG simulates a stand as a grid of these gaps, taking into account the spatial relationship between gaps when modeling light penetration. Three fundamental processes are simulated in ZELIG: tree growth, mortality, and regeneration. Each process is modeled by estimating maximum potential behavior and then constraining this potential by limiting resources such as light, soil moisture, soil fertility, and ambient temperature. Tree growth is a function of a tree’s leaf area. Mortality includes both density-dependent mortality and density- independent mortality, the latter to account for other causes of mortality such as disease, insects, or windthrow. Regeneration is determined by a species' regeneration rank relative to other species. Modifying ZELIG Parameters ZELIG.PNW v 2.0 has been parameterized and used primarily for the western hemlock zone using data from the Oregon Cascades (Urban et al. 1993, Hansen et al. 1995). For CLAMS, we modified ZELIG to better simulate forest development in the Coast Range. In addition, we developed two variants of ZELIG, one for the “fog zone” along the coast, and another for the interior Coast Range and foothills. Modifying species-specific parameters involved a three-pronged, iterative process in which ZELIG simulations were calibrated with: Calibration of ZELIG (see next panel) was based on numerous stand attributes, including basal area, tree density, quadratic mean diameter (qmd), diameter distributions, and tree mortality. 1) ORGANON simulations, for examining how ZELIG modeled even-aged stands (n=9) of Douglas-fir. 2) Long-term data sets (n=9) from Cascade Head, for evaluating ZELIG’s simulation of red alder stands or mixed-species stands such as hemlock/spruce, D-fir/hemlock, and red alder/conifers. 3) Chronosequences constructed from more than 700 stand surveys in the Coast Range, to determine if ZELIG simulations were consistent across a wide range of stand ages and conditions. References Garman, S. L., Cissel, J.H, and Mayo, J. H. (accepted) Accelerating development of late-successional conditions in young managed Douglas-fir stands: a simulation study. USDA Forest Service Pacific Northwest Research Station, PNW Gen. Tech. Report. Hann, D.W., Hester, A.S., and Olsen, C.L. 1997. ORGANON User’s Manual: Edition 6.0. Dept of Forest Resources, OSU. Hansen, A.J., Garman, S.L., Weigand, D.L., Urban, D.L., McComb, W.C., and Raphael, M.G. 1995. Alternative silvicultural regimes in the Pacific Northwest: simulations of ecological and economic effects. Ecological Applications 5: 535-554. Urban, D.L., Harman, M.E., and Halpern, C.B. 1993. Potential response of Pacific Northwestern forests to climatic change: effects of stand age and initial composition. Climatic Change 23: 247-266. Urban, D.L. and Shugart, H.H. 1992. Individual-based models of forest succession. Pages 249-292 in D.C. Glenn-Lewin, R.K. Peet, and T.T. Veblen, editors. Plant succession: theory and prediction. Chapman and Hall, London. 1) ZELIG vs. ORGANON 2) ZELIG vs. long-term data from Cascade Head Exp. For. Calibrating ZELIG (Examples of model performance with calibration data sets) 3) ZELIG vs. Coast Range chronosequences (black dots = stand survey data; colored lines = Zelig simulations of a subset of those data) 5-yr old stand with 100% Douglas-fir BA Basal area (m 2 /ha) 0 20 40 60 80 100 Density (trees/ha) 0 500 1000 Simulation time (years) 0 50 100 QMD (cm) 0 20 40 60 80 100 20-yr old stand with 95% Douglas-fir BA Simulation time (years) 0 50 100 Simulation time (years) 0 50 100 35-yr old stand with 91% Douglas-fir ORGANON (D-fir) ZELIG (D-fir) ORGANON (hemlock) ZELIG (hemlock) Sitka spruce / western hemlock stand Douglas-fir / western hemlock stand ZELIG variants for CLAMS: • Fog zone • Interior and foothills Uses of ZELIG in CLAMS ZELIG is used to simulate a variety of management practices and natural forest processes, primarily on federal and state lands. Unmanaged (“let it grow”): forest development in reserves, wilderness areas, and in managed stands prior to thinning. Response to gap formation: forest development within canopy gaps of varying sizes (0.04 ha, 0.24 ha, and 1.0 ha). Thinning strategies in young to mature stands: - single- or multi-entry heavy thinning in 20-60 year-old plantations (target densities range from 74-247 trees/ha (30-100 trees/ac)) - thin once from below to a basal area of 18.4 m 2 /ha (80 ft 2 /ac) (federal) - thin once from below to a basal area of 34.4 m 2 /ha (150ft 2 /ac) (state) - thin 3 times from below to a basal area of 28.7 m 2 /ha (125ft 2 /ac) (state) Green-tree retention: simulate 15 years of growth in new plantations under various leave-tree scenarios (modeled in ORGANON after that) Stand age (years) 0 50 100 150 200 300 400 500 Basal Area (m 2 /ha) 0 20 40 60 80 100 120 140 Stand age (years) 0 50 100 150 200 300 400 500 Trees/ha 0 200 400 600 800 1000 1200 1400 1600 1800 2000 3000 4000 Testing ZELIG (Examples of model performance with independent data sets) Simulation year = 0 Stand age = 85 0 10 20 30 40 50 60 70 80 90 100 120 140 Trees/ha 0 50 100 Douglas-fir (real) (Zelig) Hemlock (real) (Zelig) Spruce (real) (Zelig) Simulation year = 10 Stand age = 95 0 10 20 30 40 50 60 70 80 90 100 120 140 Trees/ha 0 50 100 Simulation year = 53 Stand age = 138 0 10 20 30 40 50 60 70 80 90 100 120 140 Trees/ha 0 50 100 Simulation year = 63 Stand age = 148 Dbh (cm) 0 10 20 30 40 50 60 70 80 90 100 120 140 Trees/ha 0 50 100 Simulation year = 0 Stand age = 85 0 10 20 30 40 50 60 70 80 90 100 120 140 Trees/ha 0 50 100 Douglas-fir (real) (Zelig) Hemlock (real) (Zelig) Spruce (real) (Zelig) Simulation year = 11 Stand age = 96 0 10 20 30 40 50 60 70 80 90 100 120 140 Trees/ha 0 50 Simulation year = 43 Stand age = 128 0 10 20 30 40 50 60 70 80 90 100 120 140 Trees/ha 0 50 Simulation year = 59 Stand age = 144 Dbh (cm) 0 10 20 30 40 50 60 70 80 90 100 120 140 Trees/ha 0 50 Black Rock Forest (State of Oregon) Thinned and unthinned plots --- 39 years old at simulation time 0 Siuslaw National Forest Unthinned Douglas-fir growth and yield plots --- 50 years old at time 0 Cascade Head Experimental Forest Unthinned alder/conifer plot --- 12 years old at simulation time 0 Summary of model calibration and testing Total basal area Basal area (m 2 /ha) 20 30 40 50 60 70 Quadratic mean diameter Stand age (years) 40 45 50 55 60 65 QMD (cm) 20 30 40 50 Trees > 75 cm dbh Stand age (years) 40 45 50 55 60 65 TPH 0 5 10 15 20 Thinned - ZELIG Thinned - real Not thinned - ZELIG Not thinned - real Tree density TPH 0 200 400 600 800 1000 1200 Conifer Basal Area Basal area (m 2 /ha) 50 60 70 80 90 100 110 Quadratic mean diameter Stand age (years) 60 80 100 120 140 QMD (cm) 20 30 40 50 60 Trees > 75 cm dbh Stand age (years) 60 80 100 120 140 TPH 0 10 20 30 40 50 60 Plot SI04 - ZELIG Plot SI04 - real Plot SI05 - ZELIG Plot SI05 - real Tree density TPH 200 400 600 800 1000 Hardwood basal area Basal area (m 2 /ha) 0 10 20 30 40 50 Quadratic mean diameter Stand age (years) 0 20 40 60 80 100 120 QMD (cm) 0 10 20 30 40 Trees > 75 cm dbh Stand age (years) 0 20 40 60 80 100 120 TPH 0 5 10 15 20 Plot CH23 - ZELIG Plot CH23 - real Tree density TPH 0 500 1000 1500 2000 2500 3000 Trees/ha 0 50 100 150 200 Bigleaf maple (real) Bigleaf maple (ZELIG) Douglas-fir (real) Douglas-fir (ZELIG) W. redcedar (real) W. redcedar (ZELIG) W. hemlock (real) W. hemlock (ZELIG) Stand age = 65 yrs DBH class midpoint (cm) 5 15 25 35 45 55 65 75 85 Trees/ha 0 50 100 150 200 Stand age = 85 yrs Plot SI04 - Dbh distribution at two times Plot CH23 - Dbh distribution at two times Thinned plot - Dbh distribution at two times Trees/ha 0 200 400 600 800 1000 1200 Bigleaf maple (real) Bigleaf maple (ZELIG) Douglas-fir (real) Douglas-fir (ZELIG) Red alder (real) Red alder (ZELIG) W. redcedar (real) W. redcedar (ZELIG) Sitka spruce (real) Sitka spruce (ZELIG) W. hemlock (real) W. hemlock (ZELIG) Stand age = 31 years (Zelig age = 32) DBH class midpoint (cm) 5 15 25 35 45 55 65 Trees/ha 0 50 100 150 200 250 Stand age = 71 years (Zelig age = 72) • Three data sets representing a diversity of stand ages, species compositions, stand histories, and regional climatic patterns were used to calibrate ZELIG parameters. • ZELIG simulations closely matched ORGANON’s in basal area of Douglas-fir stands. Qmd’s in ZELIG were usually higher and tree densities lower than in ORGANON, possibly because of differences in the way mortality is simulated in the two models. Tree mortality in ZELIG arises from both density-dependent and density-independent functions, resulting in higher mortality (and commensurate volumes of dead wood) than in ORGANON (data not shown). • Overall, simulations of independent data sets show that ZELIG does a good job tracking stand attributes, although discrepancies are evident in some individual stands. ZELIG tended to underestimate the rate of development of large Sitka spruce and Douglas-fir. Further adjustment of species growth parameters or temperature response functions in the model may be necessary. DBH class midpoint (cm) 5 15 25 35 45 55 65 75 Trees/ha 0 50 100 150 200 Trees/ha 0 50 100 150 200 Bigleaf maple (real) Bigleaf maple (ZELIG) Douglas-fir (real) Douglas-fir (ZELIG) Red alder (real) Red alder (ZELIG) W. redcedar (real) W. redcedar (ZELIG) W. hemlock (real) W. hemlock (ZELIG) Interior ecoregion Interior ecoregion
Transcript of Use of the ZELIG Forest Simulation Model in the Uses of ...lands, forest structure may be more...
Use of the ZELIG Forest Simulation Model in theCoastal Landscape Analysis and Modeling Study (CLAMS)
Robert J. Pabst1, Matthew N. Goslin2, Thomas A. Spies3, and Steven L. Garman1
1Department of Forest Science, OSU 2NOAA, Santa Cruz, CA 3USDA Forest Service PNW Research Station, Corvallis
IntroductionBackgroundSimulating the development and distribution of forest types across the landscape is a key component of CLAMS, requiring robust models of stand-level dynamics. Two forest growth models are being used in CLAMS. ORGANON v.6.0 (Hann et al. 1997), an empirical growth and yield model, is used to simulate stands under private and tribal land ownerships. Stands on federal and state lands are modeled with ZELIG.PNW v.2.0 (Urban et al. 1993, Garman et al., in press). ZELIG is a gap model that can simulate a wide range of stand conditions and processes, thus it should be better suited than ORGANON to modeling forests on public lands where, in comparison to privatelands, forest structure may be more complex, forest reserves exist, and alternative silviculture systems are often used. In this poster we present how we modified ZELIG for the Coast Range, how it performed with independent data sets, and its uses in CLAMS.
Model structure of ZELIGZELIG models the forest as a collection of gaps, with each gap representing the area potentially occupied by a canopy dominant tree (Urban and Shugart 1992). The model simulates each gap as it progresses through different phases of development, tracking all of the individual trees within it. ZELIG simulates a stand as a grid of these gaps, taking into account the spatial relationship between gaps when modeling light penetration.
Three fundamental processes are simulated in ZELIG: tree growth,mortality, and regeneration. Each process is modeled by estimating maximum potential behavior and then constraining this potential by limiting resources such as light, soil moisture, soil fertility, and ambient temperature. Tree growth is a function of a tree’s leaf area. Mortality includes both density-dependent mortality and density-independent mortality, the latter to account for other causes ofmortality such as disease, insects, or windthrow. Regeneration is determined by a species' regeneration rank relative to other species.
Modifying ZELIG ParametersZELIG.PNW v 2.0 has been parameterized and used primarily for the western hemlock zone using data from the Oregon Cascades (Urban et al. 1993, Hansen et al. 1995). For CLAMS, we modified ZELIG to better simulate forest development in the Coast Range. In addition, we developed two variants of ZELIG, one for the “fog zone” along the coast, and another for the interior Coast Range and foothills.
Modifying species-specific parameters involved a three-pronged, iterative process in which ZELIG simulations were calibrated with:
Calibration of ZELIG (see next panel) was based on numerous stand attributes, including basal area, tree density, quadratic mean diameter (qmd), diameter distributions, and tree mortality.
1) ORGANON simulations, for examining how ZELIG modeled even-aged stands (n=9) of Douglas-fir.
2) Long-term data sets (n=9) from Cascade Head, for evaluating ZELIG’s simulation of red alder stands or mixed-species stands such as hemlock/spruce, D-fir/hemlock, and red alder/conifers.
3) Chronosequences constructed from more than 700 stand surveys in the Coast Range, to determine if ZELIG simulations were consistent across a wide range of stand ages and conditions.
ReferencesGarman, S. L., Cissel, J.H, and Mayo, J. H. (accepted) Accelerating development of late-successional conditions in young managed Douglas-fir stands: a simulation study. USDA Forest Service Pacific Northwest Research Station, PNW Gen. Tech. Report.
Hann, D.W., Hester, A.S., and Olsen, C.L. 1997. ORGANON User’s Manual: Edition 6.0. Dept of Forest Resources, OSU.
Hansen, A.J., Garman, S.L., Weigand, D.L., Urban, D.L., McComb, W.C., and Raphael, M.G. 1995. Alternative silvicultural regimes in the Pacific Northwest: simulations of ecological and economic effects. Ecological Applications 5: 535-554.
Urban, D.L., Harman, M.E., and Halpern, C.B. 1993. Potential response of Pacific Northwestern forests to climatic change: effects of stand age and initial composition. Climatic Change 23: 247-266.
Urban, D.L. and Shugart, H.H. 1992. Individual-based models of forest succession. Pages 249-292 in D.C. Glenn-Lewin, R.K. Peet, and T.T. Veblen, editors. Plant succession: theory and prediction. Chapman and Hall, London.
1) ZELIG vs. ORGANON
2) ZELIG vs. long-term data from Cascade Head Exp. For.
Calibrating ZELIG(Examples of model performance with calibration data sets)
3) ZELIG vs. Coast Range chronosequences(black dots = stand survey data; colored lines = Zelig simulations of a subset of those data)
5-yr old stand with100% Douglas-fir BA
Bas
al a
rea
(m2 /h
a)
0
20
40
60
80
100
ORGANONZELIG
Den
sity
(tr
ees/
ha)
0
500
1000
Col 1 vs Col 5 Col 1 vs Col 8
Simulation time (years)0 50 100
QM
D (
cm)
0
20
40
60
80
100
20-yr old stand with95% Douglas-fir BA
ORGANON -- Douglas-firZELIG -- Douglas-firORGANON -- HemlockZELIG -- Hemlock
Col 1 vs Col 5 Col 1 vs Col 8 Col 1 vs Col 7 Col 1 vs Col 9
Simulation time (years)0 50 100
Simulation time (years)0 50 100
35-yr old stand with91% Douglas-fir
ORGANON (D-fir)ZELIG (D-fir)ORGANON (hemlock)ZELIG (hemlock)
Sitka spruce / western hemlock stand Douglas-fir / western hemlock stand
ZELIG variants for CLAMS:• Fog zone• Interior and foothills
Uses of ZELIG in CLAMSZELIG is used to simulate a variety of management practices and natural forest processes, primarily on federal and state lands.
Ø Unmanaged (“let it grow”): forest development in reserves, wildernessareas, and in managed stands prior to thinning.
Ø Response to gap formation: forest development withincanopy gaps of varying sizes (0.04 ha, 0.24 ha, and 1.0 ha).
ØThinning strategies in young to mature stands:- single- or multi-entry heavy thinning in 20-60 year-old plantations
(target densities range from 74-247 trees/ha (30-100 trees/ac)) - thin once from below to a basal area of 18.4 m2/ha (80 ft2/ac) (federal)- thin once from below to a basal area of 34.4 m2/ha (150ft2/ac) (state)- thin 3 times from below to a basal area of 28.7 m2/ha (125ft2/ac) (state)
Ø Green-tree retention: simulate 15 years of growth in new plantationsunder various leave-tree scenarios (modeled in ORGANON after that)
Stand age (years)0 50 100 150 200 300 400 500
Bas
al A
rea
(m2 /h
a)
0
20
40
60
80
100
120
140
Stand age (years)0 50 100 150 200 300 400 500
Tre
es/h
a
0200400600800
10001200140016001800
200030004000
Testing ZELIG(Examples of model performance with independent data sets)
Simulation year = 0Stand age = 85
0 10 20 30 40 50 60 70 80 90 100 120 140
Tre
es/h
a
0
50
100 Douglas-fir (real) (Zelig)Hemlock (real) (Zelig)Spruce (real) (Zelig)
Simulation year = 10Stand age = 95
0 10 20 30 40 50 60 70 80 90 100 120 140
Tre
es/h
a
0
50
100
Simulation year = 53Stand age = 138
0 10 20 30 40 50 60 70 80 90 100 120 140
Tre
es/h
a
0
50
100
Simulation year = 63Stand age = 148
Dbh (cm)
0 10 20 30 40 50 60 70 80 90 100 120 140
Tre
es/h
a
0
50
100
Simulation year = 0Stand age = 85
0 10 20 30 40 50 60 70 80 90 100 120 140
Tre
es/h
a
0
50
100 Douglas-fir (real) (Zelig)Hemlock (real) (Zelig)Spruce (real) (Zelig)
Simulation year = 11Stand age = 96
0 10 20 30 40 50 60 70 80 90 100 120 140
Tre
es/h
a
0
50
Simulation year = 43Stand age = 128
0 10 20 30 40 50 60 70 80 90 100 120 140
Tre
es/h
a
0
50
Simulation year = 59Stand age = 144
Dbh (cm)
0 10 20 30 40 50 60 70 80 90 100 120 140
Tre
es/h
a
0
50
Black Rock Forest (State of Oregon)Thinned and unthinned plots --- 39 years old at simulation time 0
Siuslaw National ForestUnthinned Douglas-fir growth and yield plots --- 50 years old at time 0
Cascade Head Experimental ForestUnthinned alder/conifer plot --- 12 years old at simulation time 0
Summary of model calibration and testing
Total basal area
Bas
al a
rea
(m2 /h
a)
20
30
40
50
60
70
Quadratic mean diameter
Stand age (years)40 45 50 55 60 65
QM
D (
cm)
20
30
40
50 Trees > 75 cm dbh
Stand age (years)40 45 50 55 60 65
TP
H
0
5
10
15
20
Thinned - ZELIGThinned - realNot thinned - ZELIGNot thinned - real
Tree density
TP
H
0
200
400
600
800
1000
1200
Conifer Basal AreaBas
al a
rea
(m2 /h
a)
50
60
70
80
90
100
110
Quadratic mean diameter
Stand age (years)
60 80 100 120 140
QM
D (
cm)
20
30
40
50
60 Trees > 75 cm dbh
Stand age (years)
60 80 100 120 140
TP
H
0
10
20
30
40
50
60
Plot SI04 - ZELIGPlot SI04 - realPlot SI05 - ZELIGPlot SI05 - real
Tree density
TP
H
200
400
600
800
1000
Hardwood basal area
Bas
al a
rea
(m2 /h
a)
0
10
20
30
40
50
Quadratic mean diameter
Stand age (years)0 20 40 60 80 100 120
QM
D (
cm)
0
10
20
30
40Trees > 75 cm dbh
Stand age (years)0 20 40 60 80 100 120
TP
H
0
5
10
15
20Plot CH23 - ZELIGPlot CH23 - real
Tree density
TP
H
0
500
1000
1500
2000
2500
3000
Tre
es/h
a
0
50
100
150
200
Bigleaf maple (real)Bigleaf maple (ZELIG)Douglas-fir (real)Douglas-fir (ZELIG)W. redcedar (real)W. redcedar (ZELIG)W. hemlock (real)W. hemlock (ZELIG)
Stand age = 65 yrs
DBH class midpoint (cm)5 15 25 35 45 55 65 75 85
Tre
es/h
a
0
50
100
150
200Stand age = 85 yrs
Plot SI04 - Dbh distribution at two times
Plot CH23 - Dbh distribution at two times
Thinned plot - Dbh distribution at two times
Tre
es/h
a
0
200
400
600
800
1000
1200Bigleaf maple (real)Bigleaf maple (ZELIG)Douglas-fir (real)Douglas-fir (ZELIG)Red alder (real)Red alder (ZELIG)W. redcedar (real)W. redcedar (ZELIG)Sitka spruce (real)Sitka spruce (ZELIG)W. hemlock (real)W. hemlock (ZELIG)
Stand age = 31 years (Zelig age = 32)
DBH class midpoint (cm)5 15 25 35 45 55 65
Tre
es/h
a
0
50
100
150
200
250Stand age = 71 years (Zelig age = 72)
• Three data sets representing a diversity of stand ages, species compositions, stand histories, and regional climatic patterns were used to calibrate ZELIG parameters.
• ZELIG simulations closely matched ORGANON’s in basal area of Douglas-fir stands. Qmd’s in ZELIG were usually higher and tree densities lower than in ORGANON, possibly because of differences in the way mortality is simulated in the two models. Tree mortality in ZELIG arises from both density-dependent and density-independent functions, resulting in higher mortality (and commensurate volumes of dead wood) than in ORGANON (data not shown).
• Overall, simulations of independent data sets show that ZELIG does a good job tracking stand attributes, although discrepancies are evident in some individual stands. ZELIG tended to underestimate the rate of development of large Sitka spruce and Douglas-fir. Further adjustment of species growth parameters or temperature response functions in the model may be necessary.
DBH class midpoint (cm)5 15 25 35 45 55 65 75
Tre
es/h
a
0
50
100
150
200
Tre
es/h
a
0
50
100
150
200Bigleaf maple (real)Bigleaf maple (ZELIG)Douglas-fir (real)Douglas-fir (ZELIG)Red alder (real)Red alder (ZELIG)W. redcedar (real)W. redcedar (ZELIG)W. hemlock (real)W. hemlock (ZELIG)
Interior ecoregion Interior ecoregion
